〔Abstract〕 Objective To recommend appropriate screening methods and paths for grassroots institutions and communities, so as to improve the diabetes screening ability of grassroots medical institutions or community health service centers and improve diabetes prevention and control status. Methods A total of 2083 high-risk diabetes population in 7 communities in Shenzhen Dapeng New District from April 2019 to April 2021 were screened for diabetes by point-of-care testing (POCT) glycosylated hemoglobin (HbA1c). Based on the screening results of POCT HbA1c, the diagnostic efficacy of Chinese diabetes risk score scale (referred to as ''scoring method'') and SENSIBLE non-lab model (referred to as ''model method'') were calculated. The prediabetes detection rate and diabetes detection rate were calculated by the screening results of POCT HbA1c. POCT HbA1c ≥ 5.7% was defined as positive for screening, otherwise negative, and the sensitivity, specificity, accuracy and Youdon index of the scoring method and the model method were compared. Results The proportion of males in the study population was 38.84% (809/2083), the mean age was (50.08 ± 9.64) years old, and the mean POCT HbA1c was (5.75 ± 1.01) %. Prediabetes was found in 31.83% (663/2083) and diabetes was found in 15.99% (333/2083). The sensitivity, specificity, accuracy and Yuden index were 61.24%, 40.02 %, 50.17 % and 0.013, respectively. The sensitivity, specificity, accuracy and Yuden index of the model method were 38.76 %, 60.17 %, 49.93 % and -0.011, respectively. Except for specificity, the sensitivity and accuracy of the scoring method were higher than those of the model method, and the differences were statistically significant (P < 0.05). ConclusionPOCT HbA1c is recommended to be used as one of the screening methods in eligible communities. Scoring method can also be used in areas with relatively weak medical service capacity.